Western-State Air Quality Modeling Study
(WSAQS)
Weather Research Forecast (WRF)
Winter Modeling Workplan
Prepared by:
University of North Carolina
Institute for the Environment
100 Europa Drive, Suite 490
Campus Box 1105
Chapel Hill, NC 27599-1105
ENVIRON International Corporation
773 San Marin Drive, Suite 2115
Novato, California, 94945
415-899-0700
December 2014
CONTENTS
1.0 INTRODUCTION
1.1 Background
1.2 Overview
1.3 Organization of the Modeling Workplan
2.0 MODEL SELECTION
2.1 Description of WRF
3.0 EPISODE SELECTION
3.1 Episode Selection Criteria
3.2 Episode Selection RESULTS
4.0 METEOROLOGICAL MODELING
4.1 Model Selection and Application
4.2 WRF Domain Definition
4.3 Topographic Inputs
4.4 Vegetation Type and Land Use Inputs
4.5 Atmospheric Data Inputs
4.6 Water TemPerature
4.7 Time Integration
4.8 Diffusion Options
4.9 Lateral Boundary Conditions
4.10 Top and Bottom Boundary Conditions
4.11 FDDA Data Assimilation
4.12 WRF Physics and INput Sensitivity
4.14 WRF OUTPUT VARIABLES
4.15 WRF application Methodology
5.0 METEOROLOGICAL MODEL PERFORMANCE EVALUATION
5.1 Meteorological Process Evaluation using Field Campaign Observations
5.2 Quantitative Evaluation using Surface Meteorological Observations
6.0 REFERENCES
3
3
3
4
5
5
6
6
6
10
10
10
10
10
13
13
13
13
13
13
14
14
15
16
20
21
22
26
TABLES
Table 3.1. Intense Observations Periods (IOPs) during the PCAPS in the Salt Lake
basin.
Table 4-1. 37 Vertical layer interface definition for WSAQS WRF meteorological
model simulations.
Table 4-2. Physics options used in the initial 3SAQS WRF simulation that would be
used as starting configuration for WSAQS WRF runs.
Table 4-3. Proposed WRF Wintertime Sensitivity Experiments
Table 5-1. Meteorological model performance benchmarks for simple and
complex conditions.
WSAQS
i
8
12
17
17
24
Draft WRF Winter Modeling Workplan
FIGURES
Figure 3-1. Time series of 8-hour average ozone concentrations at all monitoring
sites in the Uintah Basin, winter 2012-2013. EPA NAAQS of 75 ppb is
shown as a red dashed line.
Figure 4-1. Proposed WRF 36/12/4 km grid structure for the WSAQS
meteorological. The domains chosen are identical to those used in the
3SAQS.
Figure 4-2. Comparison of snow cover from the NAM-12km (left) and SNODAS1km (right).
Figure 4-3. Illustration of “iterative nudging” for the P-X land surface model. The
P-X uses 2-m temperature and relative humidity for indirect soil
moisture and deep soil temperature nudging.
Figure 4-4. 2-m temperature analysis used for soil nudging with the P-X land
surface model with standard 2m temperature analyses (left) and
iterative nudging (right).
Figure 5-1. Conceptual model of meteorological processes that influence cold air
pool formation and demise.
Figure 5-2. Locations of MADIS surface meteorological modeling sites with the
WRF 4-km 3SAQS modeling domain.
Figure 5-3. Example quantitative model performance evaluation display using
soccer plots that compare monthly temperature performance (colored
symbols) for the three states: Colorado (top left), Utah (top right) and
Wyoming (bottom) with the simple and complex model performance
benchmarks (rectangles).
WSAQS
ii
9
11
18
18
19
20
22
25
Draft WRF Winter Modeling Workplan
1.0 INTRODUCTION
1.1 BACKGROUND
The Western-State Air Quality Study (WSAQS) includes cooperators from U.S. Environmental Protection
Agency (EPA), United States Forest Service (USFS), Bureau of Land Management (BLM), National Park Service
(NPS), and the state air quality management agencies of Colorado, Utah, and Wyoming. The WSAQS is
intended to facilitate air resource analyses for federal and state agencies in the states of Wyoming, Colorado,
and Utah toward improved information for the public and stakeholders as a part of the National
Environmental Policy Act (NEPA) and potentially other studies. Funded by the Environmental Protection EPA,
BLM, and USFS and with in-kind support from NPS and Colorado, Utah, and Wyoming state air agencies, by
working closely with cooperators and overseeing the various agreements, the main focus of the study is on
assessing the environmental impacts of sources related to oil and gas development and production. In
particular, the cooperators will use photochemical grid models (PGMs) to quantify the impacts of proposed
oil and gas development projects within the WSAQS region on current and future air quality, including ozone
and visibility levels in the National Parks and Wilderness Areas.
The WSAQS has hired the University of North Carolina (UNC) at Chapel Hill and ENVIRON International
Corporation (ENVIRON) to develop the technical data needed to perform the modeling for WSAQS as well as
populate the Western Data Warehouse (WSDW). UNC/ENVIRON is working closely with the NPS and other
cooperators to develop technical capacity and expertise, and train NPS staff.
A key component for the WSAQS PGM modeling is the meteorological inputs. Meteorological inputs for
PGMs are generated using prognostic meteorological models. Wintertime ozone formation near oil and gas
development areas in the Rocky Mountain region is driven by a unique combination of emissions, dynamical,
and radiative conditions that are challenging to simulate in PGMs. The conventional WRF modeling
configurations used to simulate summer ozone formation do not work well for wintertime modeling. While
different groups have documented WRF configurations for simulating specific winter ozone episodes (e.g
Ahmadov et al., 2014), there has not been an effort to develop and test a wintertime WRF configuration that
yields acceptable performance across all of the Western states. This document is a workplan for developing
and testing an operational WRF wintertime configuration for simulating the meteorological conditions
conducive to ozone formation.
1.2 OVERVIEW
The pilot (2013 through September 2014) Three State Air Quality Study (3SAQS) performed meteorological
modeling for the year 2011 using the Weather Research and Forecasting Model version 3.5.1 focused on the
intermountain west states of Colorado, Utah, and Wyoming. The WRF 2011 meteorological fields were
subsequently used in photochemical grid model (PGM) application and model performance evaluation (MPE).
The MPE for 2011 identified areas of poor model performance within the study region (UNC and ENVIRON,
2014a,b). Of particular concern was the wintertime model performance in two locations in the
intermountain west. High winter ozone events have occurred in the Jonah-Pinedale Anticline Development
(JPAD) area in the Upper Green River area of southwest Wyoming and the Uinta Basin, Utah that are both
rural oil and gas development areas. Lessons learned from the prior modeling sensitivities and input from
the U.S. EPA, the University of Utah (UUtah), the Utah Department of Environmental Quality Division of Air
Quality (UTDAQ), Wyoming Department of Environmental Quality (WDEQ) and others are being used to
develop model sensitivities to improve the meteorological model performance during the winter months. In
particular, UUtah and UTDAQ have tested various WRF model configurations to improve model performance
WSAQS
3
DRAFT Winter Modeling Workplan
in simulating cold air pools in the Uintah Basin and Salt Lake Valley, which are important contributors to poor
air quality in oil and gas development regions (Neeman et al. 2014).
The inability to model or the misrepresentation of the cold air pools is a serious limitation for PGM studies
investigating the air quality impacts of natural gas and oil development in the West U.S. In particular, snow
cover is a key parameter that needs to be properly represented within a meteorological model for a number
of reasons. An increase in snow cover increases the surface albedo and near-surface actinic flux leading to
elevated photolysis rates (Schnell et al., 2009) favoring high near surface ozone concentrations. The increase
in snow cover also favors a highly stable boundary layer. These conditions generate cold air pools that may
persist for many days. Cold pool events are often associated with complexity in other synoptic and mesoscale
atmospheric circulations, in particular warming in the free atmosphere associated with warm air advection
and subsidence (Lareu et al. 2013). Other meteorological model deficiencies include sensitivity to
initialization time (personal communication Erik Crossman and Lance Avery), representation of low cloud
cover (Gultepe et al. 2014), and turbulence/mixing within the PBL. The WRF PBL schemes allow too much
turbulent mixing in the winter favoring deeper than observed boundary layers (Baklanov et al. 2011; Holstag
et al. 2013).
This document presents the meteorological modeling workplan to develop an operational winter modeling
configuration that: 1) optimizes the WRF meteorological model performance for simulating winter ozone
events across the three-state region and 2) tests the best winter model configuration for a summer month
relative to the 2011 annual 3SAQS WRF model configuration to provide guidance as to the model
configuration to use outside of the winter months. The WRF configuration improvements developed here
will be implemented in an improved 2011 air quality modeling platform, especially during the winter months,
and in a preliminary 2014 modeling platform.
1.3 ORGANIZATION OF THE MODELING WORKPLAN
The WSAQS winter meteorological modeling will leverage lessons learned from the meteorological modeling
from the 3SAQS and input from other studies and scientists working on similar issues, in particular as part of
studies of the winter ozone events in the Uinta Basin and JPAD areas. Details of the 3SAQS WRF modeling
technical approach can be found in the 2011 WRF Modeling Protocol (ENVIRON and UNC, 2013) and in the
2011 WRF Application/Evaluation report (UNC and ENVIRON, 2014a). Although the WSAQS modeling
analysis is not currently being performed to fill any particular regulatory requirement, such as a State
Implementation Plan (SIP) attainment demonstration or an Environmental Impact Statement (EIS) or
Resource Management Plan (RMP) as part of the National Environmental Policy Act (NEPA), it is being
conducted with the same level of technical rigor as a SIP-type analysis and the databases developed may
ultimately be used as a basis for regulatory air quality modeling. This WSAQS winter meteorological
modeling workplan has the following sections:
1.
2.
3.
4.
Introduction: Presents an overview and objectives of the study.
Model Selection: Introduces the model selected for the study.
Episode Selection: Describes the modeling period for the study.
Meteorological Modeling: Describes how the meteorological modeling will be conducted
including modeling domains, model options and application strategy.
5. Meteorological Model Performance Evaluation: Provides the procedures for conducting the
model performance evaluation of the meteorological model.
6. References: References cited in the document.
WSAQS
4
DRAFT Winter Modeling Workplan
2.0 MODEL SELECTION
This section discusses the selection of the Weather Research and Forecasting (WRF) meteorological model
for the WSAQS. The selection methodology follows EPA’s guidance for regulatory modeling in support of
ozone and PM2.5 attainment demonstration modeling and showing reasonable progress with visibility goals
(EPA, 2007). EPA recommends that models be selected for regulatory ozone, PM and visibility studies on a
“case-by-case” basis with appropriate consideration being given to the candidate models’:

Technical formulation, capabilities and features;

Pertinent peer-review and performance evaluation history;

Public availability; and

Demonstrated success in similar regulatory applications.
The 3SAQS modeling protocol (ENVIRON and UNC, 2013) provides an overview of WRF capabilities,
performance, and success in prior regulatory applications.
2.1 DESCRIPTION OF WRF1
The non-hydrostatic version of the Advanced Research version of the Weather Research and Forecast (WRFARW2) model (Skamarock et al. 2004; 2005; 2006; Michalakes et al. 1998; 2001; 2004) is a three-dimensional,
limited-area, primitive equation, prognostic model that has been used widely in regional air quality model
applications. The basic model has been under continuous development, improvement, testing and open
peer-review for more than 10 years. It has been used world-wide by hundreds of scientists for a variety of
mesoscale studies, including cyclogenesis, polar lows, cold-air damming, coastal fronts, severe
thunderstorms, tropical storms, subtropical easterly jets, mesoscale convective complexes, desert mixed
layers, urban-scale modeling, air quality studies, frontal weather, lake-effect snows, sea-breezes,
orographically induced flows, and operational mesoscale forecasting. WRF is a next-generation mesoscale
prognostic meteorological model routinely used for urban- and regional-scale photochemical, fine particulate
and regional haze regulatory modeling studies. Developed jointly by the National Center for Atmospheric
Research and the National Centers for Environmental Prediction, WRF is maintained and supported as a
community model by researchers and practitioners around the globe. The code supports two modes: the
Advanced Research WRF (ARW) version and the Non-hydrostatic Mesoscale Model (NMM) version. WRFARW has become the new standard model used in place of the older Mesoscale Meteorological Model
(MM5) for regulatory air quality applications in the U.S. It is suitable for use in a broad spectrum of
applications across scales ranging from hundreds of meters to thousands of kilometers.
1
2
http://www.wrf-model.org/index.php
All references to WRF in this document refer to the WRF-ARW
WSAQS
5
DRAFT Winter Modeling Workplan
3.0 EPISODE SELECTION
EPA’s ozone, PM2.5, and visibility SIP modeling guidance (EPA, 2007)3 contains recommended procedures for
selecting modeling episodes, while also referencing EPA’s 1-hour ozone modeling guidance for episode
selection (EPA, 1991)4. This Chapter presents the modeling period selected for performing the WSAQS and
the justification and rationale for its selection.
3.1 EPISODE SELECTION CRITERIA
EPA’s modeling guidance lists primary criteria for selecting episodes for ozone, PM2.5, and visibility SIP
modeling along with a set of secondary criteria that should also be considered.
3.1.1 Primary Episode Selection Criteria
EPA’s modeling guidance (EPA, 2007) identifies four specific criteria to consider when selecting episodes for
use in demonstrating attainment of the 8-hour ozone or PM2.5 NAAQS:
1. A variety of meteorological conditions should be covered, including the types of meteorological
conditions that produce 8-hour ozone and 24-hour PM2.5 exceedances in the western U.S.;
2. Choose episodes having days with monitored 8-hour daily maximum ozone and 24-hour PM2.5
concentrations close to the ozone and PM2.5 Design Values;
3. To the extent possible, the modeling data base should include days for which extensive data bases
(i.e. beyond routine aerometric and emissions monitoring) are available; and
4. Sufficient days should be available such that relative response factors (RRFs) for ozone projections
can be based on several (i.e., > 10) days with at least 5 days being the absolute minimum.
3.1.2 Secondary Criteria
EPA also lists four “other considerations” to bear in mind when choosing potential 8-hour ozone episodes
including:
1.
2.
3.
4.
Choose periods which have already been modeled;
Choose periods that are drawn from the years upon which the current Design Values are based;
Include weekend days among those chosen; and
Choose modeling periods that meet as many episode-selection criteria as possible in the maximum
number of nonattainment areas as possible.
EPA suggests that modeling an entire summer ozone season for ozone or an entire year for PM2.5 would be a
good way to assure that a variety of meteorological conditions are captured and that sufficient days are
available to construct robust relative response factors (RRFs) for the 8-hour ozone and PM2.5 Design Value
projections.
3.2 EPISODE SELECTION RESULTS
To improve the WRF modeling performance for the wintertime conditions compared to the 3SAQS
routine (i.e., summer configuration) 2011 WRF application, we selected two wintertime periods to test
sensitivities with the WRF model: December 2010 – March 2011 (includes overlap from the 3SAQS) and
December 2012 – March 2013. Extensive field measurement campaigns operated during these periods
3
4
http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rh-guidance.pdf
http://www.epa.gov/ttn/scram/guidance/guide/uamreg.pdf
WSAQS
6
DRAFT Winter Modeling Workplan
in both the Uinta Basin and JPAD region to characterize the meteorology, emissions, and the resulting
air quality, especially during cold air pool events. We did not select the December 2011-Februrary 2012
period because during this period the ozone concentrations were well below the NAAQS and cold air
pools were virtually absent in the Uintah Basin due to low snow cover (Utah DEQ, 2014). The field
campaign measurements that we propose to use to help develop and evaluate the WSAQS WRF winter
modeling configuration include, but are not limited to:
1) Persistent Cold Air Pool Study (PCAPS) within the Salt Lake basin5
2) Uinta Basin Winter Ozone Studies (UBWOS)6
3) Upper Green River Winter Ozone Study (UGRWOS)7
The measurements from these field campaigns will allow us to characterize a WRF configuration for two
different winter seasons and three different basins. The range of high ozone events covered by these studies
will provide a sufficient number of days to analyze the meteorological conditions that produce high winter
ozone. During these periods, there are many known meteorological events that contribute to both high
ozone and PM2.5. The UBWOS and UGRWOS focus on high ozone in the Uinta and JPAD regions,
respectively. The PCAPS campaign includes a focus on high PM2.5 around Salt Lake City. Below we
include a brief description of the field campaigns and select air pollution episodes captured in the
measurements.
PCAPS is a three-year project to investigate the processes leading to the formation, maintenance and
destruction of persistent cold air pools within the Salt Lake basin. The PCAPS study helped to identify
processes that lead to and the buildup, maintenance and breakup of the cold air pools, the
consequences of these processes on air pollution, and to improve meteorological models for more
accurate simulations of these events. Table 3-1 shows the Intense Observational Periods (IOPs) from
December 1, 2010 through February 7, 2011 for PCAPS. We will use the relevant observations within
these episode periods to evaluate the meteorological model performance.
UBWOS was launched to gain a better understanding of wintertime ozone formation within the Uinta
Basin. Some objectives of this field campaign included creating a long-term record of atmospheric
chemistry and meteorology, and understand the meteorological and chemical processes active in
wintertime ozone production in the Uinta Basin (Utah State, 2013). Observations are available for both
of the winter seasons that we propose to model. In particular, 2013 experienced numerous 8-hr
average ozone events that exceeded the NAAQS of 75 ppb for several days, as seen in Figure 3-1. Several
of these events are persistent and begin in January 2013. We will use information on the timing of the
high ozone events to identify meteorological episodes for model evaluation.
UGRWOS is a set of field studies initiated by the Wyoming Department of Environmental Quality
(WDEQ) to better understand baseline and elevated wintertime ozone within the JPAD region. These
field studies were initiated after the first measured high ozone values in 2005, with the area since being
designated as “non-attainment” with respect to the national ozone standard.8 The UGRWOS field
studies consist of various meteorological and air quality measurements. Of particular interest for this
workplan are observed high ozone events in early 2011. February and March 2011 saw the highest 8-hr
5
http://pcaps.utah.edu/
http://www.deq.utah.gov/NewsNotices/annualreport/Planning/s12.htm.
7
http://deq.state.wy.us/aqd/Upper%20Green%20Winter%20Ozone%20Study.asp
8
http://rd.usu.edu/files/uploads/2012_summary_ugrb.pdf
6
WSAQS
7
DRAFT Winter Modeling Workplan
average observed ozone concentrations since 2007 with 26 days exceeding 65 ppb. We will use these
high ozone events in February and March 2011 for selecting episodes to use for evaluating the WRF
winter modeling.
Table 3.1. Intense Observations Periods (IOPs) during the PCAPS in the Salt Lake basin.9
IOP Dates
IOP1
Begin (DD-MM-YYYY)
01-12-2010 12:00 UTC
End (DD-MM-YYYY)
07-12-2010 02:00 UTC
IOP2
IOP3
07-12-2010 12:00 UTC
12-12-2010 12:00 UTC
10-12-2010 15:00 UTC
14-12-2010 21:00 UTC
IOP4
24-12-2010 12:00 UTC
26-12-2010 21:00 UTC
IOP5
IOP6
IOP7
IOP8
01-01-2011 00:00 UTC
11-01-2011 12:00 UTC
20-01-2011 12:00 UTC
23-01-2011 12:00 UTC
09-01-2011 12:00 UTC
17-01-2011 20:00 UTC
22-01-2011 06:00 UTC
26-01-2011 12:00 UTC
IOP9
26-01-2011 12:00 UTC
31-01-2011 06:00 UTC
IOP10
02-02-2011 18:00 UTC
05-02-2011 18:00 UTC
9
Comment
Long lived cold pool.
PCAP began on 29 Nov,
before PCAPS
operations began
Short and sweet."
Followed by great
troughiness
"Short and sweet."
again. The Christmas
Cold Pool
Likely a biggie
Somewhat uneventful,
observations were
minimal
Very intensive
operations, sidewall
and canyon ops
Intensive lake based
operations
IOPs obtained from http://pcaps.utah.edu/about/status.php
WSAQS
8
DRAFT Winter Modeling Workplan
Figure,1
Figure
3-1. 1.,
TimeTime,series,of,8
series of 8-hourhour,average,ozone,concentrations,at,all,monitoring,sites,in,
average ozone concentrations at all monitoring sites in the
Uintah
Basin, winter
2012-2013.13.,,EPA,NAAQS,of,75,ppb,is,shown,as,a,red,dashed,line.,
EPA NAAQS of 75 ppb is shown as a red dashed line.10
the,Uintah,Basin
,winter,2012
"
Figure,1 2., Time,series,of,average,snow,depth,from,five,stations,in,the,Uintah,Basin ,
pseudo lapse,rate,for,the,Basin ,8 hour,average,ozone,at,Ouray ,and,average,total,daytime,
UV A,and,UV B,radiation,(average,during,daytime,hours,of,the,sum,of,upwelling,and,
downwelling,UV A,and,UV B),at,Horsepool ,winter,2012 13.,,The,pseudo lapse,rate,was,
derived,from,the,change,in,temperature,with,elevation,at,surface,meteorological,stations,in,
the,Basin.,,The,dashed,black,line,indicates,a,lapse,rate,of,zero,and,an,ozone,concentration,of,
75,ppb.,,A,more,negative,lapse,rate,indicates,a,stronger,inversion.,
10
Figure 3-1 from http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013-FinalReport/UBOS_2013Sec_3_DistribMon.pdf
6"
WSAQS
9
DRAFT Winter Modeling Workplan
4.0 METEOROLOGICAL MODELING
The WRF meteorological model will be applied for the winter months of 2010-2011 and 2012-2013 using a
36/12/4 km domain structure. However, cold air pools can be forced by small-scale processes (e.g.
turbulence; Laureau and Horel, 2014) and may require higher resolution simulations. We propose adding a
higher-resolution modeling domain for a single configuration and episode if the 4-km simulation fails to
improve the representation of cold air pool events. The higher resolution simulation will include a 1.33-km
domain covering the oil and gas basins in northeast Utah, southwest Wyoming, and northern Colorado.
4.1 MODEL SELECTION AND APPLICATION
The latest publicly available version of WRF (version 3.6.1 released August 14, 2014) will be used in the
WSAQS meteorological modeling. The WRF preprocessor programs including GEOGRID, UNGRIB, and
METGRID will be used to develop model inputs.
4.2 WRF DOMAIN DEFINITION
The WRF computational grid was designed so that it can generate CAMx/CMAQ meteorological inputs for the
same 36 km CONUS , 12 km western US, and 4 km three state domains as used in the 3SAQS. The WRF
modeling domain was defined to be slightly larger than the CAMx/CMAQ PGM modeling domains to
eliminate boundary artifacts in the meteorological fields used as input to CAMx/CMAQ. Such boundary
artifacts can occur as the boundary conditions (BCs) for the meteorological variables come into dynamic
balance with WRF’s atmospheric equations and numerical methods for downscaling from coarser to finerresolution domains. Figure 4-1 depicts the WRF horizontal modeling domains to be used in the WSAQS. The
outer 36 km continental U.S. (CONUS) domain (D01) has 165 x 129 grid cells, selected to be consistent with
the existing Regional Planning Organization (RPO) and EPA modeling CONUS domain. The projection is
Lambert Conformal with the “national RPO” grid projection center of 40o, -97o with true latitudes of 33o and
45o. The 12 km western U.S. domain has 256 × 253 grid cells with offsets from the 36 km grid of 15 columns
and 26 rows. The 4 km three state domain has 301 x 361 grid cells with offsets from the 12 km grid of 75
columns and 55 rows.
The WRF model will use the same 37 vertical layers sigma-pressure surfaces as defined in the 3SAQS, Table 41. This vertical layer configuration considers approximately the same number of vertical levels in the lowest
200 meters as in other meteorological models studies resolving cold air pools (i.e. Neemann et al. 2014).
4.3 TOPOGRAPHIC INPUTS
Topographic information for the WRF modeling will be developed using the standard WRF terrain
databases available from the National Center for Atmospheric Research (NCAR).11 The 36 km CONUS
domain was based on the 10 min. (~18 km) global data. The 12 km western US domain will be based on
the 2 min. (~4 km) data. The 4 km three state domain will be based on the 30 sec. (~900 m) data.
4.4 VEGETATION TYPE AND LAND USE INPUTS
Vegetation type and land use information will be derived from two different land use databases: the
USGS 24-category and NLCD 2006 40-category databases. These land use databases are distributed with
WRF version 3.6.1 and are proposed for consistency with other WRF modeling efforts within other
groups, including the U.S. EPA and the UUtah. These efforts are described in more detail in Section 4.12.
11
http://dss.ucar.edu/
WSAQS
10
DRAFT Winter Modeling Workplan
The standard WRF surface characteristics corresponding to each land use category will be employed
with the exception of snow cover. For snow, the albedo will be adjusted for each land use cover type.
This adjustment is important because the surface albedo can depend on vegetation type/coverage. For
example, conifer forests have a lower snow albedo because they consist of dark trees covering the high
albedo snowpack. Grasslands have a high snow albedo because they could be completely covered by
the snowpack. This level of detail can have important implications for the surface energy budget and
dependent issues.
Figure 4-1. Proposed WRF 36/12/4 km grid structure for the WSAQS meteorological. The
domains chosen are identical to those used in the 3SAQS.
WSAQS
11
DRAFT Winter Modeling Workplan
Table 4-1. 37 Vertical layer interface definition for WSAQS WRF meteorological model
simulations.
WRF Meteorological Model
WRF
Layer
37
36
35
34
33
32
31
30
29
28
27
26
25
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
WSAQS
Sigma
0.0000
0.0270
0.0600
0.1000
0.1500
0.2000
0.2500
0.3000
0.3500
0.4000
0.4500
0.5000
0.5500
0.6000
0.6400
0.6800
0.7200
0.7600
0.8000
0.8400
0.8700
0.8900
0.9100
0.9300
0.9400
0.9500
0.9600
0.9700
0.9800
0.9850
0.9880
0.9910
0.9930
0.9950
0.9970
0.9985
1.0000
Pressure
(mb)
50.00
75.65
107.00
145.00
192.50
240.00
287.50
335.00
382.50
430.00
477.50
525.00
572.50
620.00
658.00
696.00
734.00
772.00
810.00
848.00
876.50
895.50
914.50
933.50
943.00
952.50
962.00
971.50
981.00
985.75
988.60
991.45
993.35
995.25
997.15
998.58
1000
12
Height
(m)
19260
17205
15355
13630
11930
10541
9360
8328
7408
6576
5816
5115
4463
3854
3393
2954
2533
2130
1742
1369
1098
921
747
577
492
409
326
243
162
121
97
72
56
40
24
12
0
Thickness
(m)
2055
1850
1725
1701
1389
1181
1032
920
832
760
701
652
609
461
440
421
403
388
373
271
177
174
171
84
84
83
82
82
41
24
24
16
16
16
12
12
DRAFT Winter Modeling Workplan
4.5 ATMOSPHERIC DATA INPUTS
The WRF initialization fields will be taken from the 12 km (Grid #218) North American Model (NAM) archives
available from the National Climatic Data Center (NCDC) NOMADS server.
4.6 WATER TEMPERATURE
The water temperature data will be taken from the NCEP RTG global one-twelfth degree analysis.12
4.7 TIME INTEGRATION
Third-order Runge-Kutta integration will be used (rk_ord = 3). The maximum time step, defined for the
outer-most domain (36 km) only, should be set by evaluating the following equation:
𝑑𝑡 =
6𝑑𝑥
𝐹𝑚𝑎𝑝
Where dx is the grid cell size in km, Fmap is the maximum map factor (which can be found in the output from
REAL.EXE), and dt is the resulting time-step in seconds. For the case of the 36 km RPO domain, dx = 36 and
Fmap = 1.08, so dt should be taken to be less than 200 seconds. Longer time steps risk CFL errors, associated
with large values of vertical velocity, which tend to occur in areas of steep terrain (especially during very
stable conditions typical of winter).
For the current modeling, adaptive time stepping will be used to decrease total runtime for the annual
simulation. We suggest using target_cfl = 0.8, starting_time_step = -1, and min_time_step = -1 (for each
domain). The max_time_step = 180, 60, 20 should be used, with the default values for
max_step_increase_pct. Any initializations that fail due to CFL errors should be re-run with
use_adaptive_time_step = .false., with time_step set to 180 or 120 seconds.
4.8 DIFFUSION OPTIONS
Horizontal Smagorinsky first-order closure (km_opt = 4) with sixth-order numerical diffusion and
suppressed up-gradient diffusion (diff_6th_opt = 2) will be used.
4.9 LATERAL BOUNDARY CONDITIONS
Lateral boundary conditions will be specified from the initialization dataset on the 36 km domain with
continuous updates nested from the 36 km domain to the 12 km domain and from the 12 km domain to the
4 km domain, using one-way nesting (feedback = 0).
4.10 TOP AND BOTTOM BOUNDARY CONDITIONS
The top boundary condition will be selected as an implicit Rayleigh dampening for the vertical velocity.
Consistent with the model application for non-idealized cases, the bottom boundary condition will be
selected as physical, not free-slip.
12
http://polar.ncep.noaa.gov/sst
WSAQS
13
DRAFT Winter Modeling Workplan
4.11 FDDA DATA ASSIMILATION
The WRF model will be run with a combination of analysis and observational nudging (i.e., Four Dimensional
Data assimilation [FDDA]). Analysis nudging will be used on the 36 km and 12 km domain. For winds and
temperature, analysis nudging coefficients of 5x10-4 and 3.0x10-4 will be used on the 36 km and 12 km
domains, respectively. For mixing ratio, an analysis nudging coefficient of 1.0x10-5 will be used for both the
36 km and 12 km domains. The nudging uses both surface and aloft nudging with no nudging for
temperature and mixing ratio in the lower atmosphere (i.e., within the boundary layer). Observational
nudging will be performed on the 4 km grid domain using the Meteorological Assimilation Data Ingest System
(MADIS)13 observation archive. The MADIS archive includes the National Climatic Data Center (NCDC)14
observations and the National Data Buoy Center (NDBC) Coastal-Marine Automated Network C-MAN15
stations. The observational nudging coefficients for winds and temperatures will be 1.2x10-4, 6.0x10-4,
respectively and the radius of influence will be set to 60 km.
4.12 WRF PHYSICS AND INPUT SENSITIVITY
The WRF sensitivities proposed here are based on knowledge gained from the prior 3SAQS meteorological
modeling, recent studies working to improve WRF model performance during the wintertime over complex
terrain, and advice from meteorological modelers working on similar problems. The preliminary WRF
wintertime configuration will use the same physics as in the final 3SAQS configuration but with the latest
release of WRF (v3.6.1). Table 4-2 lists the 3SAQS WRF physics options to be used as a base configuration for
the WSAQS wintertime modeling. Our goal is to improve upon the 3SAQS WRF model configuration,
especially during the wintertime. The WRF configurations will be run and evaluated for two full winter
seasons.
Snow cover is known to be an important component of wintertime ozone events and that improving its
representation in the model improves the simulation of cold air pools. We will directly use the Snow Data
Assimilation System (SNODAS16, Barrett 2003) product for snow cover in WRF. SNODAS integrates a
physically based, spatially distributed energy and mass balance model with observed snow data from satellite
and airborne platforms, and ground stations. SNODAS output has high spatial resolution (~1 km) and
temporal resolution (1 hour) over the United States. Our implementation of SNODAS in WRF will include
adjustments to the snow albedo that are based on land use type. SNODAS will replace the NAM snow cover
used in the previous 3SAQS WRF configuration. Figure 4-2 illustrates that there are large differences in snow
cover between NAM and SNODAS.
The 2008 and 2011 3SAQS WRF simulations used the Noah land surface model (LSM), which is consistent
with prior WRF simulations conducted by UUtah and UTDAQ. When WRF is run for operational air quality
simulations, it is typically reinitialized every 5.5 days. In particular, the Noah LSM in WRF includes a snow
sub-model that is initialized using a snow cover database (i.e. from NAM or SNODAS) and reinitializing the
snow in addition to the atmospheric fields is important. Discussions with UUtah and UTDAQ provided insight
into a potential problem with the timing and frequency of reinitializing in WRF. In particular, initializing the
model in the middle of a cold air pool event may create problems when trying to represent these events.
This is a difficult operational problem given timing of these events and that they are likely happening at
different times throughout the domain. Reinitializing during a cold pool event is a problem that may be
avoided by carefully constructing the reinitialization time for known events or by using the Pleim-Xiu (P-X)
13
Meteorological Assimilation Data Ingest System. http://madis.noaa.gov/
National Climatic Data Center. http://lwf.ncdc.noaa.gov/oa/ncdc.html
15 National Data Buoy Center. http://www.ndbc.noaa.gov/cman.php
14
16
Snow Data Assimilation System. http://nsidc.org/data
WSAQS
14
DRAFT Winter Modeling Workplan
land surface model (LSM). The P-X LSM directly assimilates SNODAS data, thus avoiding reinitialization. If
reinitialization happens during a known cold pool and associated poor air quality event, we will run at least
one of these cold pool events continuously to evaluate the impacts of reinitializing the model during a cold
pooling episode.
The P-X LSM is a well-known option in WRF to drive retrospective meteorological simulations for regulatory
air quality applications (Gilliam and Pleim, 2010). As the P-X LSM does not include a sub-snow model, it can
directly assimilate the SNODAS dataset using FDDA. WRF modeling by the U.S. EPA no longer reinitialize the
model every 5.5 days (per conversation Rob Gilliam, US EPA) and they have recently demonstrated significant
model performance improvements for near surface fields when recycling the WRF model 2-m temperature
as a background field to Obsgrid for soil nudging (Gilliam et al. 2014). Figures 4-3 and 4-4 and illustrate this
“iterative nudging” approach with an example of differences between the iterative and standard nudging
techniques. We will test the impacts of continuous integration, directly assimilating SNODAS, and iterative
nudging with a WRF sensitivity run for both winter seasons. Note that this P-X sensitivity test will use the
NLCD 2006 LULC data and the Asymmetric Convective Model (ACM version 2; Gilliam and Pleim, 2010). We
will use the ACM PBL scheme for consistency with ongoing efforts at the U.S. EPA. Additionally, WRF PBL
sensitivity tests performed by both the U.S. EPA and UTDAQ have shown have shown little improvement in
near surface fields relative to the magnitude of observed errors (per conversation Chris Misenis, U.S. EPA and
Lance Avey, UTDAQ).
Table 4-3 lays out the WRF configuration experiments that we will conduct to optimize the performance of
the model for winter conditions. The first configuration will use the Noah LSM with USGS LULC data. The
novel winter configuration that we will test will initialize the Noah LSM using SNODAS and include land-use
specific albedo adjustments. For this configuration we will develop a special data processor to format the
SNODAS data for input to the WRF-REAL program. This configuration will reinitialize every 5.5 days as in the
prior 3SAQS, which includes reinitializing back to SNODAS.
In the second configuration we will use the P-X LSM with NCLD LULC data and the ACM2 PBL scheme. WRFREAL will also be used to create SNODAS snow-cover inputs to WRF, which will be assimilated directly into PX. This simulation will not reinitialize but rather use a continuous integration initialized on December 1st.
We will build on the second configuration and apply an iterative nudging approach for the final WRF
configuration experiment. All three experiments will be run for the two winter season episodes described
above and evaluated against the available field campaign meteorology observations. A possible fourth
sensitivity test includes a high-resolution 1.33-km WRF simulation for a single configuration and episode.
This simulation will only be performed if the prior 4-km simulations fail to improve the representation of cold
air pool events, especially for near surface fields. The higher resolution simulation will include a 1.33-km
covering the oil and gas basins in northeast Utah, southwest Wyoming, and northern Colorado.
The wintertime configuration chosen with the smallest overall error within the Uinta, Upper Green River, and
Salt Lake basins will be used to simulate July of 2011. This additional test will help determine if different
winter and summer configurations are necessary for simulating meteorology to support air quality
assessments in the Western U.S.
4.14 WRF OUTPUT VARIABLES
The WRF model will be configured to output additional variables in order to support air quality modeling with
the Community Multiscale Air Quality Model (CMAQ). The following fields will be activated in the WRF
output history files for the 3SAQS: fractional land use (LANDUSEF), aerodynamic resistance (RA), stomatal
WSAQS
15
DRAFT Winter Modeling Workplan
resistance (RS), vegetation fraction in the Pleim-Xiu LSM (VEGF_PX), roughness length (ZNT), inverse MoninObukhov length (RMOL)
4.15 WRF APPLICATION METHODOLOGY
The WRF model when using the Noah land surface option will be executed in 5-day blocks initialized at 12Z
every 5.5 days. Twelve (12) hours of spin-up will be included in each 5-day block before the data will be used
in the subsequent evaluation. The WRF model when using the P-X land surface option will be executed with
a continuous integration with 12 hours of spin-up. Model results will be output every 60 minutes and output
files will be split at 12-hour intervals. Additional experiments will include testing using a longer runtime given
the length of known cold air pool events. WRF will be configured to run in distributed memory parallel mode.
WSAQS
16
DRAFT Winter Modeling Workplan
Table 4-2. Physics options used in the initial 3SAQS WRF simulation that would be used as
starting configuration for WSAQS WRF runs.
WRF Treatment
Option Selected
Notes
Microphysics
Thompson
Longwave Radiation
RRTMG
Shortwave Radiation
RRTMG
Land Surface Model (LSM)
NOAH
Two-layer scheme with
vegetation and sub-grid tiling.
Planetary Boundary Layer (PBL)
scheme
YSU
Cumulus parameterization
Kain-Fritsch in the 36 km and
12 km domains. None in the 4
km domain.
Nudging applied to winds,
temperature and moisture in
the 36 km and 12 km domains
Yonsie University (Korea)
Asymmetric Convective Model
with non-local upward mixing
and local downward mixing.
4 km can explicitly simulate
cumulus convection so
parameterization not needed.
Temperature and moisture
nudged above PBL only
Analysis nudging
A scheme with ice, snow, and
graupel processes suitable for
high-resolution simulations.
Rapid Radiative Transfer Model
for GCMs includes random
cloud overlap and improved
efficiency over RRTM.
Same as above, but for
shortwave radiation.
Observation Nudging
Nudging applied to surface
wind and temperature only in
the 4 km domain
Initialization Dataset
12 km North American Model
(NAM)
moisture observation nudging
produces excessive rainfall
Table 4-3. Proposed WRF Wintertime Sensitivity Experiments
Winter
Season
Dec. 2010 –
Mar. 2011
Dec. 2012 –
Mar. 2013
2011 Episode
17
18
Exp 1. SNODAS
w/ Albedo
Adjustments
Exp. 2.
Exp 3. P-X Land
Reinitialization Surface Model
NA
Noah LSM with
USGS17 LULC data
and SNODAS
initialization
N/A
P-X LSM &
ACM2 PBL with
NLCD 200618
LULC data and
assimilated
SNODAS
Same as Exp 2.
with iterative
nudging of soil
temperature and
moisture
N/A
N/A
NA
Exp. 1
Exp 4. Iterative
Nudging
Exp. 5 High
Resolution
N/A
Exp. 1, 2, or 3
USGS Land Cover Database. http://landcover.usgs.gov/
National Land Cover Database 2006. http://www.mrlc.gov/nlcd2006.php
WSAQS
17
DRAFT Winter Modeling Workplan
Figure 4-2. Comparison of snow cover from the NAM-12km (left) and SNODAS-1km (right).19
Met Analysis (NAM)
Obs. Data
OBSGRID
WRF analysis (4-km)
4-km WRF
WRF output (4-km)
Figure 4-3. Illustration of “iterative nudging” for the P-X land surface model. The P-X uses 2m temperature and relative humidity for indirect soil moisture and deep soil temperature
nudging.
19
Images courtesy of Rob Gilliam, US EPA.
WSAQS
18
DRAFT Winter Modeling Workplan
Figure 4-4. 2-m temperature analysis used for soil nudging with the P-X land surface
model with standard 2m temperature analyses (left) and iterative nudging (right).
WSAQS
19
DRAFT Winter Modeling Workplan
5.0 METEOROLOGICAL MODEL PERFORMANCE EVALUATION
The WRF model evaluation approach will primarily use a process-oriented evaluation. The WRF evaluation
will focus on time and locations with high wintertime ozone. Figure 5-1 is a conceptual model of known
meteorological processes important for wintertime ozone. The simulated meteorological processes seen in
Figure 5-1 will be compared against field campaign data such as those obtained from PCAPS, UBWOS, and
UGRWOS. The following are some of the meteorological processes to be evaluated, where possible:

synoptic scale processes such as the preconditioning of cold air near the surface, warm air advection
aloft, and large-scale atmospheric subsidence.

mesoscale processes such as the spillover of warm air into the valley creating dynamic instability and
thermally driven upslope and downslope flows.

surface and boundary layer processes such as the amount of snow cover and/or cloud cover,
temperature inversions, boundary layer heights, wind speed and direction, and net solar radiation.
Below we discuss the data used for the process oriented evaluation and additional quantitative evaluation of
near surface meteorological observations.
Figure 5-1. Conceptual model of meteorological processes that influence cold air pool
formation and demise.
WSAQS
20
DRAFT Winter Modeling Workplan
5.1 METEOROLOGICAL PROCESS EVALUATION USING FIELD CAMPAIGN OBSERVATIONS
A number of software tools including R20, GRADS21, NCL22, and AMET23 will be used to compare field
campaign measurements and observations at the surface and aloft. Below is a list of the types of the
observational datasets and the meteorological processes being evaluated.










915 Wind Profiler with RASS (Radio Acoustic Sounding System) and 449 Wind Profiler
o Vertical profiles of temperature
o Vertical profiles of wind speed
Ceilometer – aerosol optical depth
o Clouds
o Relative humidity
Mobile soundings from glider
o Temperature
o Relative Humidity
HOBO mobile temperature logger
Surface meteorological tower sites
o Winds
o Temperature
o Relative humidity
o Barometric Pressure
o Solar radiation
o Soil Parameters
SODAR/Sonic anemometer
o Boundary layer height
Doppler Lidar
o Vertical profiles of winds
Surface meteorological data (ASOS, MESONET sites, field sites)
o Temperature
o Relative Humidity
o Winds
Upper-air soundings (RAOBs)
o Vertical profiles of temperature
o Vertical profile of dewpoint
o Vertical profile of winds
SNOTEL data
o Snow depth, air temperature, precipitation
20
http://www.r-project.org/
http://iges.org/grads/
22
http://www.ncl.ucar.edu/
23
http://www.cmascenter.org
21
WSAQS
21
DRAFT Winter Modeling Workplan
Overall, the meteorological data from the field studies will be used to evaluate the meteorological processes
important for simulating the timing, strength, and length of the cold air pool episodes important for
wintertime ozone.
5.2 QUANTITATIVE EVALUATION USING SURFACE METEOROLOGICAL OBSERVATIONS
The statistical evaluation approach examines tabulations and graphical displays of the model bias and error
for surface wind speed, wind direction, temperature, and mixing ratio (humidity) and compares the
performance statistics to benchmarks developed based on a history of meteorological modeling as well as
past meteorological model performance evaluations. Interpretation of bulk statistics over a continental or
regional scale domain is problematic, and it is difficult to detect if the model is missing important subregional features. For this analysis the statistics will be performed on the three states of Colorado, Wyoming,
and Utah and select valley basins within the 4-km domain, with additional domain-wide basis for the 36 km
CONUS and 12 km modeling domains. The WRF evaluation procedures used by 3SAQS study will be used in
the WSAQS (ENVIRON and UNC, 201324).
The observed database for winds, temperature, and water mixing ratio to be used in this analysis are the
National Oceanic and Atmospheric Administration (NOAA) Earth System Research Laboratory (ESRL)
Meteorological Assimilation Data Ingest System (MADIS). The locations of the MADIS monitoring sites within
the 4 km 3SAQS WRF modeling domains are shown in Figure 5-2.
Figure 5-2. Locations of MADIS surface meteorological modeling sites with the WRF 4-km
3SAQS modeling domain.
24
http://vibe.cira.colostate.edu/wiki/Attachments/Modeling/3SAQS_2011_WRF_MPE_v8_draft_Aug04_2014.pdf
WSAQS
22
DRAFT Winter Modeling Workplan
The quantitative model performance evaluation of WRF using surface meteorological measurements will
be performed using the publicly available METSTAT25 and AMET26 evaluation tools. Both tools calculate
statistical performance metrics for bias, error, and correlation for surface winds, temperature, and
mixing ratio and can produce time series of predicted and observed meteorological variables and
performance statistics. A full annual model evaluation is very difficult to summarize in a single
document, especially a simulation that could be used for many different purposes. With this in mind,
the WRF model evaluation will present results for several subregions and even at the individual site level
within the three-state area, leaving potential data users to independently judge the adequacy of the
model simulation. Overall comparisons are offered to judge the model efficacy, but this review does not
necessarily cover all potential user needs and applications.
Statistical metrics will be presented for each state, for each RPO, and for the United States portion of
the 36 km, 12 km, and 4 km modeling domains. To evaluate the performance of the WRF 2011
simulation for the U.S., a number of performance benchmarks for comparison will be used. Emery et al.
2001 derived and proposed a set of daily performance “benchmarks” for typical meteorological model
performance. These standards were based upon the evaluation of about 30 MM5 and RAMS
meteorological simulations of limited duration (multi-day episodes) in support of air quality modeling
study applications performed over several years. These were primary ozone model applications for
cities in the eastern and Midwestern U.S. and Texas that were primarily simple (flat) terrain and simple
(stationary high pressure causing stagnation) meteorological conditions. More recently these
benchmarks have been used in annual meteorological modeling studies that include areas with complex
terrain and more complicated meteorological conditions; therefore, they must be viewed as being
applied as guidelines and not bright-line numbers. That is, the purpose of these benchmarks is not to
give a passing or failing grade to any one particular meteorological model application, but rather to put
its results into the proper context of other models and meteorological data sets. Recognizing that these
simple conditions benchmarks may not be appropriate for more complex conditions, McNally (2009)
analyzed multiple annual runs that included complex terrain conditions and suggested an alternative set
of benchmarks for temperature under more complex conditions. As part of the WRAP meteorological
modeling of the western U.S., including the Rocky Mountain Region, as well as the complex conditions
up in Alaska, Kemball-Cook (2005) also came up with meteorological model performance benchmarks
for complex conditions. Table 5-1 lists the meteorological model performance benchmarks for simple
(Emery et al., 2001) and complex conditions, where we have adopted the complex benchmarks from
Kemball-Cook (2005) since they covered more of the meteorological parameters.
The key to the benchmarks is to understand how well the model performs in this case, relative to other
model applications run for the U.S. These benchmarks include bias and error in temperature, wind
direction and mixing ratio as well as the wind speed bias and Root Mean Squared Error (RMSE) between
the models and databases. The benchmark for each variable to judge whether predictions from a
meteorological model are on par with previous meteorological modeling studies are as follows:
26
http://www.cmascenter.org
WSAQS
23
DRAFT Winter Modeling Workplan
Table 5-1. Meteorological model performance benchmarks for simple and complex
conditions.
Parameter
Temperature Bias
Temperature Error
Mixing Ratio Bias
Mixing Ratio Error
Wind Speed Bias
Wind Speed RMSE
Wind Direction Bias
Wind Direction Error
Simple
≤ ±0.5 K
≤ 2.0 K
≤ ±1.0 g/kg
≤ 2.0 g/kg
≤ ±0.5 m/s
≤ 2.0 m/s
≤ ±10 degrees
≤ 30 degrees
Complex
≤ ±2.0 K
≤ 3.5 K
NA
NA
≤ ±1.5 m/s
≤ 2.5 m/s
NA
≤ 55 degrees
The equations for bias and error are given below, with the equation for the Root Mean Squared Error (RMSE)
similar to the unsigned error metric only being the square of the differences between the prediction and
observation and a square root is taken of the entire quantity.
Bias =
Error =
N
 P  O 
1
N
i 1
1
N
i
i
N
 P O
i 1
i
i
Figure 5-3 displays an example model performance soccer plot graphic from the 3SAQS 2011 WRF run for
monthly temperature performance within the 4 km domains. Shown in the “soccer plot” is the monthly
temperature bias (x-axis) versus error (y-axis) as colored symbols with the simple and complex performance
benchmarks27 represented by the rectangles. When the WRF monthly performance achieves the benchmark,
it falls within the rectangle (i.e., scores a goal). In this example we see the 2011 WRF simulation for Colorado,
Wyoming, and Utah demonstrates that nearly all of the near surface temperatures results fall within the
complex benchmark with exception of Utah during the winter. However, during the wintertime model error
increases.
27
Note that Figure 5-3 is using the McNally (2009) versions of the complex benchmark for temperature whereas
we have adopted the Kemball-Cook (2005) versions for the temperature and wind benchmarks.
WSAQS
24
DRAFT Winter Modeling Workplan
Figure 5-3. Example quantitative model performance evaluation display using soccer plots
that compare monthly temperature performance (colored symbols) for the three states:
Colorado (top left), Utah (top right) and Wyoming (bottom) with the simple and complex
model performance benchmarks (rectangles).
WSAQS
25
DRAFT Winter Modeling Workplan
6.0 REFERENCES
Ahmadov, R., et al. 2014. Understanding high wintertime ozone pollution events in an oil and
natural gas producing region of the western US. Atmospheric Chemistry Physics, 14,
20295-20343.
Baklanov, A.A., B. Grisogono, R. Bornstein, L. Mahrt, S.S. Zilitinkevich, P. Taylor, S.E. Larsen,
W.M. Rotach, and H.J.S. Fernando. 2011. The nature, theory, and modeling of
atmospheric boundary layers, Bulletin of the American Meteorological Society, 92, 123128, doi: 10.1175/2010BAMS2791.1.
Barrett, A. 2003. National Operational Hydrologic Remote Sensing Center Snow Data
Assimilation System (SNODAS) Products at NSIDC. NSIDC Special Report 11. Boulder,
CO.
Emery, C., E. Tai, and G. Yarwood, 2001. Enhanced Meteorological Modeling and Performance
Evaluation for Two Texas Ozone Episodes. Prepared for the Texas Natural Resource
Conservation Commission, prepared by ENVIRON International Corporation, Novato, CA.
31-August.
(http://www.tceq.texas.gov/assets/public/implementation/air/am/contracts/reports/m
m/EnhancedMetModelingAndPerformanceEvaluation.pdf).
ENVIRON and UNC . 2013. Three-State Air Quality Study (3SAQS): Weather Research and
Forecasting Model Draft Modeling Protocol. ENVIRON International Corporation,
Novato, California. University of North Carolina, Chapel Hill, NC. May 2013. (Add link).
UNC and ENVIRON. 2014a. Three-State Air Quality Study (3SAQS): Weather Research and
Forecasting Model – Year 2011 Modeling Application/Evaluation. University of North
Carolina, Chapel Hill, NC. ENVIRON International Corporation, Novato, California.
August 2014.
(http://vibe.cira.colostate.edu/wiki/Attachments/Modeling/3SAQS_2011_WRF_MPE_v8
_draft_Aug04_2014.pdf).
UNC and ENVIRON. 2014b. Three-State Air Quality Study (3SAQS): CAMx Photochemical Grid
Model Draft Model Performance Evaluation – Simulation year 2011. University of North
Carolina, Chapel Hill, NC. ENVIRON International Corporation, Novato, California.
November 10.
(http://views.cira.colostate.edu/tsdw/Demo/Documents.aspx).
EPA. 1991. Guideline on the Regulatory Application of the Urban Airshed Model. U.S.
Environmental Protection Agency, Office of Air Quality Planning and Standards,
technical Support Division, Source Receptor Analysis Branch, Research Triangle Park,
North Carolina. July. (http://www.epa.gov/ttn/scram/guidance/guide/uamreg.pdf).
EPA. 2007. Guidance on the Use of Models and Other Analyses for Demonstrating Attainment
of Air Quality Goals for Ozone, PM2.5 and Regional Haze. U.S. Environmental Protection
Agency, Research Triangle Park, NC. EPA-454/B-07-002. April.
(http://www.epa.gov/ttn/scram/guidance/guide/final-03-pm-rh-guidance.pdf).
WSAQS
26
DRAFT Winter Modeling Workplan
Gilliam, R., J. Pleim, W. Appel, C. Misenis, K. Baker. 2014. Improving WRF for Fine-scale Air
Quality Applications. Atmospheric Modeling and Analysis Division Seminar, March 26.
Gilliam. R.C., and J.E. Pleim. 2010. Performance Assessment of New Land Surface boundary
Layer Physics in the WRF-ARW. Journal of Applied Meteorology and Climatology, 49,
760-774. doi:http://dx.doi.org/10.1175/2009JAMC2126.1.
Gultepe, I., T. Kuhn, M. Pavolonis, C. Calvert, J. Gurka, A.J. Heymsfield, P.S. K. Liu, B. Zhou, R.
Ware, B. Ferrier, J. Milbrandt, B. Bernstein. 2014. Ice fog in arctic during FRAM-ICE fog
project: aviation and nowcasting applications. Bulletin of the American Meteorological
Society, 211-226. doi: 1175/BAMS-D-11-00071.1
Holtslag, A.A.M. G. Svensson, P. Baas, S. Basu, B. Beare, A.C.M. Beljaars, F.C. Bosveld, J. Cuxart, J.
Lindvall, G.J. Steenveld, M. Tjernstrom, B.J.H. Van De Wiel. 2013. Stable atmospheric
boundary layers and diurnal cycles: challenges for weather and climate models. Bulletin
of the American Meteorological Society. 94, 1691-1706. doi:10.1175/BAMS-D-1100187.1.
Kemball-Cook, S., Y. Jia, C. Emery and R. Morris. 2005. Alaska MM5 Modeling for the 2002
Annual Period to Support Visibility Modeling. Prepared for Western Regional Air
Partnership (WRAP). Prepared by Environ International Corporation. September.
http://pah.cert.ucr.edu/aqm/308/docs/alaska/Alaska_MM5_DraftReport_Sept05.pdf
Laureau, N.P., E.T. Crosman, C.D. Whiteman, J.D. Horel, S.W. Hoch, W.O.UJ. Brown, T.W. Horst.
2013. The persistent cold-air pool study, Bulletin of the American Meteorological
Society, 94, 51-63. doi:10.1175/BAMS-D-11-00255.1.
Laureau, N. and J. Horel, 2014. Turbulent Erosion of Persistent Cold Air Pools: Numerical
Simulations. Journal of Atmospheric Science. doi:10.1175/JAS-D-140173.1, in press.
McNally, D. E., 2009. 12km MM5 Performance Goals. Presentation to the Ad-Hoc Meteorology
Group. 25-June. http://www.epa.gov/scram001/adhoc/mcnally2009.pdf
Michalakes, J., J. Dudhia, D. Gill, J. Klemp and W. Skamarock. 1998. Design of a NextGeneration Regional Weather Research and Forecast Model. Mesoscale and Microscale
Meteorological Division, National Center for Atmospheric Research, Boulder, CO.
(http://www.mcs.anl.gov/~michalak/ecmwf98/final.html).
Michalakes, J., S. Chen, J. Dudhia, L. Hart, J. Klemp, J. Middlecoff and W. Skamarock. 2001.
Development of a Next-Generation Regional Weather Research and Forecast Model.
Developments in Teracomputing: Proceedings of the 9th ECMWF Workshop on the Use
of High Performance Computing in Meteorology. Eds. Walter Zwieflhofer and Norbet
Kreitz. World Scientific, Singapore. Pp. 269-276.
(http://www.mmm.ucar.edu/mm5/mpp/ecmwf01.htm).
Michalakes, J., J. Dudhia, D. Gill, T. Henderson, J. Klemp, W. Skamarock and W. Wang. 2004.
The Weather Research and Forecast Model: Software Architecture and Performance.
WSAQS
27
DRAFT Winter Modeling Workplan
Proceedings of the 11th ECMWF Workshop on the Use of High Performance Computing
in Meteorology. October 25-29, 2005, Reading UK. Ed. George Mozdzynski.
(http://wrf-model.org/wrfadmin/docs/ecmwf_2004.pdf).
Neeman, E.M., E.T. Crosman, J.D. Horel, L. Avey. 2014. Simulations of a cold-air pool associated
with elevated wintertime ozone in the Uintah Basin, Utah. Atmospheric Chemistry and
Physics. 15, 15953-16000. doi:10.5194/acpd-14-15953-2014
Schnell, R.C. S.J. Oltmans, R.R. Neely, M.S. Endres, J.V. Molenar, A.B. White. 2009. Rapid
photochemical production of ozone at high concentrations in a rural site during winter,
Nat. Geosci., 2, 120-122. Doi:10.1038/NGEO415
Skamarock, W. C. 2004. Evaluating Mesoscale NWP Models Using Kinetic Energy Spectra. Mon.
Wea. Rev., Volume 132, pp. 3019-3032. December.
(http://www.mmm.ucar.edu/individual/skamarock/spectra_mwr_2004.pdf).
Skamarock, W. C., J. B. Klemp, J. Dudhia, D. O. Gill, D. M. Barker, W. Wang and J. G. Powers.
2005. A Description of the Advanced Research WRF Version 2. National Center for
Atmospheric Research (NCAR), Boulder, CO. June.
(http://www.mmm.ucar.edu/wrf/users/docs/arw_v2.pdf).
Skamarock, W. C. 2006. Positive-Definite and Monotonic Limiters for Unrestricted-Time-Step
Transport Schemes. Mon. Wea. Rev., Volume 134, pp. 2241-2242. June.
(http://www.mmm.ucar.edu/individual/skamarock/advect3d_mwr.pdf).
Utah State, 2013. 2012 Uintah Basin Winter Ozone Study and Air Quality Study Final Report.
Utah State University. Feb.
(http://rd.usu.edu/files/uploads/ubos_2011-12_final_report.pdf).
Utah DEQ. 2014. Uintah Basin Ozone Study Final Report – Distributed Measurements of Air
Quality and Meteorology. Utah Department of Environmental Quality, (DEQ), Division
of Air Quality (DAQ).
(http://www.deq.utah.gov/locations/U/uintahbasin/docs/2014/03Mar/UBOS-2013Final-Report/UBOS_2013Sec_3_DistribMon.pdf).
WSAQS
28
DRAFT Winter Modeling Workplan